Intrinsic FGF2 and FGF5 promotes angiogenesis of human aortic endothelial cells in 3D microfluidic angiogenesis system

Abstract

The human body contains different endothelial cell types and differences in their angiogenic potential are poorly understood. We compared the functional angiogenic ability of human aortic endothelial cells (HAECs) and human umbilical vein endothelial cells (HUVECs) using a three-dimensional (3D) microfluidic cell culture system. HAECs and HUVECs exhibited similar cellular characteristics in a 2D culture system; however, in the 3D microfluidic angiogenesis system, HAECs exhibited stronger angiogenic potential than HUVECs. Interestingly, the expression level of fibroblast growth factor (FGF)2 and FGF5 under vascular endothelial growth factor (VEGF)-A stimulation was significantly higher in HAECs than in HUVECs. Moreover, small interfering RNA-mediated knockdown of FGF2 and FGF5 more significantly attenuated vascular sprouting induced from HAECs than HUVECs. Our results suggest that HAECs have greater angiogenic potential through FGF2 and FGF5 upregulation and could be a compatible endothelial cell type to achieve robust angiogenesis.

P

,* ,* , , ,

Neo-angiogenesis is an essential process to enhance vessel regeneration1,2. Many studies have focused on endothelial cells to explore the novel mechanisms underlying angiogenesis35. Conventionally, human umbilical vein endothelial cells (HUVECs) and human aortic endothelial cells (HAECs) are representative endothelial cell types isolated from human blood vessels, and both cell types show similar cellular characteristics and morphology6.

However, the dierences in functional characteristics between HUVECs and HAECs have not yet been fully dened.

Considering their dierent cellular origins, HUVECs and HAECs could have dierent cellular characteristics and several studies have suggested that endothelial cells have their own transcriptional and phenotypic characteristics depending on origin. For instance, the orphan nuclear receptor COUP-TFII is specically expressed in the venous endothelium and a mutation in COUP-TFII leads to the activation of arterial surface antigen in veins7.

Notch ligands and receptors are known to be expressed dierently in HUVECs and HAECs8.

Angiogenesis-related growth factors such as vascular endothelial growth factor (VEGF) or broblast growth factor (FGF) are known as important regulators of angiogenesis. During the in vivo vascular sprouting process, VEGF induces the polarization of endothelial cells and contributes to the determination of tip cell formation9.

Simultaneously, Notch signaling converts adjacent cells to stalk cells, leading to VEGF receptor expression10,11.

FGF has also been reported as involved in angiogenesis through loss-of-function studies. Previous studies suggested that the migratory response induced by FGF2 stimulation was distinct in dierent endothelial cell types; however, FGF2 represented a mild eect on the major guiding cue12. Mice lacking individual FGFs revealed a variety of phenotypes, ranging from early embryonic lethality to mild defects1315, suggesting that FGFs act in a developmental stage-specic manner. In addition, FGF ligands or their unique expression patterns in specic tissues determine the possibility of endothelial cell protrusion. FGF2 deciency in endothelial cells causes defects in endothelial cell integrity16,17, and FGF2 enhances endothelial cell proliferation and vessel repair in injured

* These authors

SCIENTIFIC REPORTS

1

www.nature.com/scientificreports/

vessels18,19. FGF5 is well known to have tight connection with hair growth cycle20, and gene transfer of FGF5 into injured myocardium was reported to promote blood ow and enhanced vessel formation21,22. However role of FGF5 for angiogenesis has not been known much. The role of FGF ligands and receptors in dierent endothelial cell types is also poorly understood.

Recently, three-dimensional (3D) microfluidic angiogenesis systems have been adopted in vascular research2325. They can form 3D tube-like angiogenic structures, perfectly circular and randomly distributed in 3D extracellular matrix (ECM) scaold. They have advantage of mimicking in vivo-like microenvironments of chemical gradients and physical stiness in the scaold, and also provide in situ quantitative analysis on the angiogenic morphology under various stimuli2628. In this study, the features of the 3D microuidic angiogenesis system were successfully adopted by in vivo mimicking of vascular sprouting via a VEGF-A gradient29 and a precise computational simulation25 to a detailed comparison of the angiogenic potential of HAECs and HUVECs.

We compared the cellular characteristics of HUVECs and HAECs in a 2D culture system. Both cell types showed a similar endothelial cell-specic cobblestone appearance (Fig.1a). Immunouorescence images show that CD31, CD144 and vWF were ubiquitously expressed in both cell types (Fig.1b). Bromodeoxyuridine (BrdU) incorporation rate was also similar between the HUVECs and HAECs (Fig.1c,d). Results from scratched wound-healing assays also showed similar in vitro wound closure rates (Fig.1e,f). Both HUVECs and HAECs showed a similar in vitro network formation, which was maintained up to 72 hours on Matrigel without any morphological dierences (Fig.1gi and Supplementary Fig. 1).

In the 2D culture systems, HUVECs and HAECs showed similar angiogenic appearances, which were veried by 3D microuidic angiogenesis system. VEGF-A (50ng/ml) was supplied to the side channel and VEGF-A (20ng/ml) was added to the two expansion channels (ECs) to generate a VEGF-A gradient in the collagen scaold. Endothelial cells (2 106 cells/ml) were seeded into the two ECs (Fig.2a). The diusion prole, estimated using COMSOL, conrmed that the VEGF-A concentration gradually decreased according to the position of all channels (Fig.2b). In the 3D microuidic angiogenesis system, the endothelial cells induced new sprouts that invaded into the scaold in a VEGF-A gradient dependent manner (Fig.2c,d and Supplementary Fig. 2). Both cell types did not possess angiogenic potential in the absence of VEGF-A; however, the vascular density of HAECs was 6.09-fold higher at 20-20 ng/ml VEGF-A and 3.27-fold higher at 50-20 ng/ml VEGF-A than that of the HUVECs. Lumen formation was observed in both endothelial cells, with very small dierences in lumen diameter between HUVECs and HAECs (Supplementary Fig. 3). Interestingly, immunouorescence staining with ZO-1 and F-actin showed a 1.52-fold higher tip cell number and a 1.95-fold longer lopodia length, respectively, in HAECs relative to HUVECs (Fig.2eg). The higher angiogenic potential of HAECs (compared to HUVECs) was maintained under hypoxic conditions of 5% and 10% oxygen. Interestingly angiogenic potential of both cell types were the highest in mild hypoxic condition of 10% oxygen, than the other two conditions of 5% and 20% oxygen (Supplementary Fig. 4).

The correlation of type I collagen gel stiness and collagen ber diameter was assessed by adjusting collagen solution pH during polymerization, conrming that smaller diameter of collagen ber in higher pH increased gel stiness30. The diameter of individual collagen ber can be measured by electron microscopy (Fig.3a). Invasion of tubular structures of both HAECs and HUVECs into the sti collagen gel was found to be signicantly reduced. However, the higher angiogenic potential of HAECs than HUVECs was maintained in both environments with high and low stiness (Fig.3b,c). The protruding region area and perimeter of cellular invasion was approximately 2.73-fold and 1.79-fold higher, respectively, in HAECs relative to HUVECs. These results demonstrate that angiogenic potential of HAECs is higher than HUVECs in both sti and so type I collagen gel (Supplementary Fig. 5). When the collagen gel was exchanged with the laminin-enriched Matrigel (growth factor reduced), HAECs successfully sprouted into the Matrigel where HUVECs failed (Fig.3d,e). These data suggest that in the 3D microuidic angiogenesis system, HAECs have stronger angiogenic potential that is independent of gel stiness and components.

Microarray analysis was performed to identify dierentially expressed transcripts in HAECs and HUVECs. HUVECs and HAECs had similar transcriptional expression proles with 96% accordance (Supplementary Fig. 6a). Factors showing two-fold increase were selected among angiogenesis-related genes using the GO term process (Fig.4a and Supplementary Fig. 6b). The angiogenesis-related genes selected by microarray analysis were veried by quantitative real-time RT-PCR. GBX2 (1.68-fold), FGF2 (2.38-fold), FGF5 (10.92-fold), and COL8A1 (2.27-fold) were upregulated and ID1 (0.62-fold), TSPAN12 (1.46-fold), CCL2 (1.27-fold), PTGS2 (0.44-fold), APOLD1 (0.53-fold), ANGPT2 (4.63-fold), and HOXA5 (1.19-fold) were downregulated in HAECs, which was dierent from the expression prole in HUVECs (Fig.4b). FGF2 and FGF5 mRNAs were most highly expressed in HAECs relative to HUVECs. FGF2 protein level was signicantly higher in both HAEC lysate (2.88-fold) and HAEC-conditioned medium (61.4-fold) relative to HUVEC, as determined by angiogenesis cytokine arrays (Fig.4c,d and Supplementary Fig. 7a,b). Semi-quantitative and quantitative real-time RT-PCR of FGF ligands and receptors showed that FGF2, FGF4, FGF5, and FGFR1 were detected in both HAECs and HUVECs; however, FGF2 and FGF5 expression was 1.59-fold and 1.23-fold higher, respectively, and FGFR1 expression was 2.31-fold lower in HAECs relative to HUVECs (Fig.4e,f).

SCIENTIFIC REPORTS

2

www.nature.com/scientificreports/

Figure 1. HUVECs and HAECs have similar cellular characteristics in a two-dimensional culture dish.

(a) Representative phase contrast and F-actin (red) immunouorescence images of HUVECs and HAECsin a 2D culture dish. Nucleus were stained with DAPI (blue). Scale bar = 100 m. (b) Endothelial cell characterization using immunouorescence staining with primary antibody followed by CD31 (green), CD144 (red), vWF (red), and DAPI (blue) staining. Scale bar=50m. (c) BrdU-incorporated HUVECs and HAECs were measured using ow cytometry. (d) Quantication of G0/G1-phase, S-phase or G2-phase cell percentage in HUVECs and HAECs. n=3, n.s.; non-signicant. (e) A phase contrast image of migrated cells observed by the time-dependent wound healing assay. Scale bar=200m. (f) Quantication of cell invasion rate against time. n=5. (g) Phase contrast images obtained during network formation assay from 024hr. Scale bar=50m. (h) Quantication of branching points per visual eld. n=4. (i) Quantication of network length per visual eld. n=4. n.s.: non-signicant.

Next, we compared the mRNA expression levels of FGF ligands and receptors in HAECs and HUVECs during the in vitro sprouting and invasion processes in the 3D microuidic angiogenesis system. Quantitative real-time RT-PCR showed that FGF2 (1.56-fold) and FGF5 (110.15-fold) were strikingly upregulated in HAECs relative to HUVECs in the 3D micro-uidic angiogenesis system (Fig.5a). Furthermore, we observed that FGF2 recombinant protein treated endothelial cells signicantly increased vascular density; however, when treated with the FGFR inhibitor SU5402, HAECs sprouting with elongated morphology and prominent filopodia were impaired (Supplementary Fig. 8ad). Finally, we compared the eect of FGF2 and FGF5 on HAECs and HUVECs sprouting angiogenic potential using FGF2 small interfering RNA (siRNA) or FGF5 siRNA in the 3D microuidic angiogenesis system. Transient transfection of FGF2 siRNA dramatically abolished the vascular invasion of both HUVECs and HAECs into ECM scaold; interestingly, FGF5 siRNA transfection signicantly reduced the angiogenic sprouting of HAECs over that of HUVECs (Fig.5b,c). Expression of phospho-broblast growth factor receptor 1 (p-FGFR1) increased in HAECs rather than HUVECs when treated with recombinant FGF2 and FGF5 (Supplementary Fig. 9ac). Intrinsic FGF2 and FGF5 in HAECs seemed to increase angiogenic potential via FGFR phosphorylation.

SCIENTIFIC REPORTS

3

www.nature.com/scientificreports/

Figure 2. HAECs have more angiogenic potential than HUVECs under VEGF-A stimulation in a 3D microuidic angiogenesis system. (a) Schematic view of the 3D microuidic angiogenesis system. (b) Simulation of the VEGF-A gradient during culture in the 3D microuidic angiogenesis system. (c) Representative phase contrast images of HUVECs and HAECs stimulated with VEGF-A at day 3. Scale bar=150m. (d) Measurement of vascular density per eld in HUVECs and HAECs. n=6. (e) Immunouorescence images of HUVECs and HAECs in the 3D microuidic angiogenesis system obtained by incubating the cells with antibody followedby ZO-1 (green), F-actin (red), and DAPI (blue). Scale bar=150m. White arrows indicate dierent lopodia extensions of tip cells. (f) Measurement of tip cell number in HUVECs and HAECs. n=6. (g) Measurement of lopodia length (m) of HUVECs and HAECs (per eld). n=6. *p<0.05, **p<0.01 and ***p<0.001 versus

HUVECs, and #p< 0.05 versus 20-20ng/ml VEGF-A group.

The study presents that; (i) HAECs show higher angiogenic potential than HUVECs under VEGF-A stimulation only in 3D microuidic angiogenesis system, and (ii) although endogenous FGF2 and FGF5 expression in both cell types is a crucial regulator of angiogenesis, increased expression of FGF5 rather than that of FGF2 extends greater angiogenic potential to HAECs.

The 3D microuidic angiogenesis system used in the study is found to be powerful for understanding stereoscopic cellular morphogenesis of 3D cooperative migration and morphogenesis of endothelial cells31,32. For

example, the regulation of type 1 collagen solution pH during gelation is known to alter the gel stiness33 in

cross-linking of the bers34. In accordance with the previous report, the collagen stiness was observed to inuence angiogenic potential of both HUVECs and HAECs (Fig.3b,c). ECM-related physical interaction might be a regulator of angiogenic potential. When endothelial cells reside in a complex 3D architecture, they perceive cues from the ECM through cell-cell or chemo-mechanical coupling, and nally generate angiogenic responses35. In the present study, both cell types have similar cellular characteristics and function in 2D culture systems; however, HAECs demonstrated signicantly higher angiogenic potential than HUVECs into the type I collagen gel in the 3D microuidic angiogenesis system. This suggests that an intrinsic factor in both cell types might play a role in angiogenic process.

SCIENTIFIC REPORTS

4

www.nature.com/scientificreports/

Figure 3. HAECs have greater angiogenic potential than HUVECs owing to dierences in collagen ber stiness and porosity. (a) Electron microscopy images of the eect of pH on collagen ber stiness. Scale bar=1m. (b,c) Phase contrast images of collagen ber stiness in the 3D microuidic angiogenesis system and quantication of vascular density. Scale bar=150m. (d,e) Phase contrast and immunouorescence images of the vascular density of HUVECs and HAECs lled with laminin-enriched gel obtained by incubating the cells with antibody followed by F-actin (red) and DAPI (blue) staining. Scale bar =150m. *p<0.05, **p<0.01 and ***p< 0.001 versus HUVEC group.

Some angiogenesis-related factors were known to show dierent expression patterns in each endothelial cell type. The VEGF-VEGFR signaling pathway is known to regulate vessel formation for both arterial and venous endothelial cells36,37. Moreover, F11R protein, known as a crucial indicator for atherosclerosis, was induced in both HUVECs and HAECs aer treatment with the same concentration of TNF and INF38. On the other hand, VCAM-1 expression in HAECs was signicantly increased by treatment with a 20-fold higher concentration of sCD40L than that used in HUVECs39. EphrinB2 was also highly expressed in HAECs, whereas EphB4 was only expressed in HUVECs but not in HAECs40. In addition, hypoxia potentiated agonist evoked responses in arterial endothelial cells but not in venous endothelial cells41. These data suggest that several angiogenesis-related factors show dierent expression patterns in each endothelial cell type. Microarray experiments in the present study observed that GBX2 and COL8A were highly expressed in HAECs, but ID1, TSPAN12, CCL2, PTGS2, APOLD1, ANGPT2, and HOXA5 were highly expressed in HUVECs. They had a good correlation with the previous studies. In dierent types of endothelial cells, pharyngeal arch artery development was involved in the upregulation of GBX242. COL8A was also reported to be associated with VEGF-A in angiogenesis-dependent macular degeneration43, whereas ANGPT2, a regulatory factor for vessel remodeling44, was highly expressed in HUVECs relative to HAECs. These results suggest that the angiogenic potential of HUVECs and HAECs are inuenced by dierent growth factors and cytokines, and they depend on cell origin. In addition, these results suggest that GBX2, COL8A, and ANGPT2 may be crucial factors for determining the diversity of specialized endothelial cell types.

Cytokines and growth factors secreted from endothelial cells also determine endothelial cell characteristics in an autocrine manner. In the angiogenesis cytokine array performed on cell lysates, EGF, FGF2, uPAR, TIMP1, and TIMP2 are highly expressed in HAEC lysates relative to HUVEC lysates. Interestingly, FGF2 is only expressed in HAEC-conditioned medium, but not in HUVEC-conditioned medium, suggesting that FGF2 might play an important role in arterial endothelial cell function. In contrast, IL-8 and MMP are highly expressed in HUVECs

SCIENTIFIC REPORTS

5

www.nature.com/scientificreports/

Figure 4. FGF2 and FGF5 are highly expressed in HAECs relative to HUVECs. (a) Microarray analysis of the angiogenesis-related mRNA expression level of HUVECs and HAECs gated for two-fold higher expressed genes. n=3. (b) Quantitative PCR to conrm the mRNA expression level of HUVECs and HAECs. Values are the average of three independent experiments; they are normalized to the expression levels in HUVECs. n=3. *p<0.05, **p<0.01 and ***p< 0.001 versus HUVECs. (c) Angiogenesis cytokine array of cell lysates and conditioned medium for both cell types. (d) Quantication of FGF2 protein levels in the cell lysate and conditioned medium n=3. *p<0.05 and **p< 0.01 versus HUVECs group or HUVEC_CM group. (e) Determination of endogenous

FGF ligand and receptor levels in HUVECs and HAECs using semi-quantitative RT-PCR. (f) Quantitative PCR was performed as a conrmatory test. mRNA expression relative to the mean values observed in HUVECs. n=3. n.d.: non-detection. *p<0.05 and **p< 0.01 and versus HUVECs group.

and ANGPT2 and MCP-1 are highly expressed in both HUVEC lysates and HUVEC-conditioned medium, indicating that these molecules could play a role in venous endothelial cell specication.

FGFs are small polypeptide growth factors that contain signal peptides for secretion to the extracellular environment to promote endothelial cell growth and movement45. Four FGF receptors and 22 FGF ligand members

SCIENTIFIC REPORTS

6

www.nature.com/scientificreports/

Figure 5. FGF2 and FGF5 are crucial factors for angiogenesis in the 3D microuidic angiogenesis system.

(a) Quantitative PCR analysis was performed to conrm mRNA expression levels. mRNA expression relative to the mean values of GAPDH. n=3. *p<0.05, **p<0.01 and ***p< 0.001 versus HUVECs group. (b,c) siRNA-mediated knockdown of FGF2 and FGF5 in HUVECs and HAECs. Scale bar=150m. n=5. *p<0.05 versus HUVECs, and #p< 0.05 versus control siRNA group.

have already been reported46. FGF signaling has been implicated in many physiological and pathological processes in angiogenesis. VEGF in vertebrate endothelial cells is a fundamental regulator of vasculogenesis, angiogenesis, and lymphangiogenesis in conjunction with multiple other growth factors such as PDGF-BB, TGF-1, FGF2, S1P, ANGPT1, and ANGPT2, as well as the signaling pathways involving NOTCH and Ephrin28. One of the most interesting things about VEGF is that it acts as a critical morphogen connected by FGF2 and primarily acts as a mitogen47. Consistent with a previous report, we found the signicance of FGF2 and FGF5 during angiogenesis between HUVECs and HAECs under a VEGF-A gradient, especially when the mRNA expression of FGF5 was excessively increased in HAECs. This phenomenon was veried in the 3D microuidic angiogenesis system. Although the absolute expression level of FGF2 was higher than that of FGF5, the relative fold-change value of FGF5 was higher than that of FGF2 in HAECs. Therefore, FGF5 expression suggests a target as a sprouted endothelial cell marker.

To validate the angiogenic eect of FGF2 and FGF5, we performed various functional assays such as the siRNA technique and FGF inhibitor assessments. First, we conrmed whether FGF2 and FGF5 siRNA could inhibit the formation of sprouted structures. The formation of HAECs vessel-like structures in 3D microuidic angiogenesis system was signicantly inhibited by FGF2 and FGF5 siRNA. Our study demonstrates that the simultaneous treatment of HAECs with FGF2 or FGF5 siRNA impaired cell sprouting from a polygonal to an elongated morphology with prominent lopodia; however, the vascular density inhibited by FGF5 siRNA was not signicant in HUVECs. FGF signaling inhibitors such as SU5402 have been eectively used for the functional inhibition of FGFR48. Similarly, we observed that SU5402 impaired cell sprouting from a polygonal to an elongated morphology with prominent lopodia (Supplementary Fig. 8c,d). Similarly, we observed that SU5402 impaired cell sprouting from a polygonal to an elongated morphology with prominent lopodia (Supplementary Fig. 8c,d). The previous report suggested that FGF2 mainly regulated angiogenesis at development stage and vascular integrity49. In this study, the FGF2 was found to be crucial for angiogenesis in both HAECs and HUVECs, but FGF5 was involved in angiogenesis and vessel patterning only in HAECs under 3D microenvironment. Our

SCIENTIFIC REPORTS

7

www.nature.com/scientificreports/

understanding of the angiogenesis mechanism, where FGF2 and FGF5 expression contributes to angiogenesis, is still limited. Although a direct correlation among VEGF-A, FGF2, and FGF5 were not fully addressed, but both FGF2 and FGF5 were found to be critical to induce angiogenesis in the presence of VEGF-A.

In conclusion, the study provides evidence that FGF2 and FGF5 are strongly and selectively expressed in HAECs than HUVECs. The 3D microuidic angiogenesis system helped to apparently characterize morphological features in angiogenic procedure of the endothelial cells, which could not be dierentiated in previous 2D assays. HAECs were found to have higher angiogenic potential through the upregulation of FGF2 and FGF5, which could be a therapeutic targets for anti-angiogenic strategy.

HUVECs (BioBud Inc. Seoul, Korea) and HAECs (LONZA Walkersville Inc. Basel, Switzerland) were cultured according to the manufacturers instructions. The culture medium was composed of EGM-2 MV (Contained with 20 ng/ml VEGF-A, insulin growth factor, epidermal growth factor, bro-blast growth factor 2, hydrocortisone, ascorbic acid and RA-1000) supplemented with 5% fetal bovine serum, 100U/ml penicillin, and 50U/ml streptomycin in a fully humidied atmosphere of 5% CO2 at 37C. The medium was changed every two days.

Polydimethylsiloxane (Sylgard 184, Dow Chemicals)-based 3D microuidic angiogenesis systems were used as previously described25. Briey, polydimethylsiloxane chips were bonded with a coverslip by plasma treatment (Femto Science) and immediately coated with poly-L-lysine (P8920, Sigma) solution, followed by incubation at 37C for 4hrs. Aer coating, the chip was washed and dried at 80C for 24hrs. Type 1 collagen gel (354236, BD Bioscience) was prepared at the desired pH and concentration, injected into the gel, and incubated at 37C for 30min for gelation. Aer gelation, the channels were lled with growth medium and the cell channels were lled with the cell suspension (1.2105 cells/channel). Aer seeding, all chips were incubated at 37 C in a fully humidied atmosphere of 5% CO2, and the medium was replaced daily. Human recombinant FGF2 (AF-233, R&D Systems) and SU5402 (572630, Calbiochem) were added into the upper EC channel. Cell migration was monitored daily and images were captured using MetaMorph soware (Molecular Devices). Vascular density, capillary forming region perimeter and Capillary forming region area were analyzed with ImageJ soware (NIH Image, Bethesda, MD). Vascular density= (Total gel area Day n gel area)/ Total gel area.

Angiogenic molecules supplied in the 3D microuidic angiogenesis system were simulated as previously described25. The VEGF-A (100-20, PeproTech) gradient and the sprouting angiogenic potential of endothelial cells under a 50-20-20ng/ml VEGF-A gradient over 24h were estimated using the nite element method-based computational uid dynamics code in COMSOL Multiphysics 4.0 (COMSOL Inc.). Diusion of VEGF-A was simulated using Ficks second law: (C)/ (t) + r3(Drc) = 0, where C is VEGF-A concentration (mol/m3), and D is the diusion coefficient (m2/s). VEGF-A concentration was determined subject to the initial conditions of passive VEGF-A supply (all channels were lled with the control medium; 20-20-20ng/ml VEGF-A containing EGM-2MV) or active VEGF-A supply (a high concentration was applied to the le channel; 50-20-20ng/ml VEGF-A containing EGM-2MV).

For uorescent staining, cells were xed with 4% paraformaldehyde (P6148, Sigma) and blocked with 4% bovine serum albumin (A9418, Sigma) in phosphate-buered saline containing 0.1% Triton X-100 (PBST; X-100, Sigma) at room temperature for 1hr. The cells were incubated overnight in a humid chamber at 4C with primary antibodies: anti-human CD31 antibody (M082301, DAKO) and anti-human ZO-1 antibody (40-2300, Invitrogen), anti-human vWF antibody (A0082, DAKO), and anti-human CD144 (555661, BD). The cells were washed with PBST three times, followed by 1 hr incubation with secondary antibodies: Alexa Fluor 488 goat anti-rabbit IgG (A11008, Invitrogen), Alexa Fluor 594 goat anti-rabbit IgG (A11012, Invitrogen), Alexa Fluor 594 goat anti-mouse IgG (A11005, Invitrogen) and Rhodamine Phalloidin antibody (R415, Invitrogen). Aer staining the nucleus with 4,6-diamidino-2-phenylindole (DAPI; D1306, Invitrogen), the cells were mounted with a uorescent mounting medium (S3023, DAKO). Immunohistochemistry images were acquired using uorescence microscope (BX61, Olympus) and Zeiss LSM700 confocal uorescence microscope (Carl Zeiss). Immunouorescence staining images in 3D microuidic angiogenesis system were presented by merging of z-stack confocal images.

Cell proliferation was analyzed using the fluorescence-activated cell sorting (FACS) BrdU ow kit (559619, BD Bioscience) according to the manufacturers instructions. Briey, BrdU was added to cells and incubated in a CO2 incubator at 37 C for 1 h. The cells were xed and permeabilized with a buer (554722, BD Bioscience) for 20min. Aer incubation with 30g of DNase at 37 C for 1 h, the cells were washed twice and stained with uorescein isothiocyanate (FITC)-conjugated anti-BrdU antibody at room temperature for 20min. Aer resuspension with 7-AAD-containing buer, the cells were nally analyzed using ow cytometry (FACS Calibur, BD Bioscience).

HUVECs and HAECs were seeded into 6-well plates at a density of 4105 cells per well. Aer 24h of incubation, conuent cell monolayers were scratched across the midline using a pipette tip, and then scanned at 0, 6, 9, and 12h using an optical microscope (Optika).

SCIENTIFIC REPORTS

8

www.nature.com/scientificreports/

In vitro With about 250l of growth factor reduced BD Matrigel (356231, BD Bioscience) was evenly spread on a 24-well culture plate and allowed to solidify at 37C. HUVECs and HAECs were spread on a Matrigel-treated wells at a density of 4 104 cells per well. The cells were observed microscopically (Optika) and recorded for up to 120hrs. Phase contrast images were obtained using an Optika visual program. Network length and branching point were quantied with ImageJ soware (NIH Image, Bethesda, MD)50. Branching point indicates where network meets more than three. Network length was measured from one branching point to the end of another branching point.

The fabrication of the 3D microuidic angiogenesis system for cellular experiment was performed by monitoring the invasion of HUVECs and HAECs under normoxia and hypoxia. Both of endothelial cells were harvested by trypsinization and seeded middle and under channel in 3D microuidic angiogenesis system to a nal cell density of 1.2105 cells/channel containing 50-20-20 ng/ml VEGF gradient. 21% O2, 5% CO2 and 74% N2 was supplied incubator to create normoxic condition. Hypoxia was then generated by switching O2 gas mixture from 21% to 10% or 5%. 10% O2, 5% CO2 and 85% N2 was supplied incubator to create mild hypoxic condition. In addition, 5% O2, 5% CO2 and 90% N2 was supplied incubator to create harsh hypoxic condition. Phase contrast images were acquired at day 3 aer plating on 3D microuidic angiogenesis system.

Total RNA was extracted from HUVECs and HAECs using Trizol reagent (TR-118, MRC) according to the manufacturers recommendations. About 500ng of RNA was reverse-transcribed into complementary DNA (cDNA) using M-MLV reverse transcriptase (28025-013, Invitrogen). Polymerase chain reaction using iQTM SYBR Green supermix (170-8880, Bio-Rad) and the indicated primers was performed using the AB PCR system (Applied Biosystems) or MYiQ2 detection system (Bio-Rad). GAPDH (Forward: 5-ACCACCATGGAGAAGGC-3, Reverse: 5-GGCATGGACTGTGGTCATGA-3) was used as an internal standard. Primer sequences and product sizes are presented in Supplemental Table 1.

For microarray analysis, total RNA was extracted using the RNeasy Plus Mini kit (74134, Qiagen). Microarray fabrication was conducted at the BML Corporation using their proprietary technology. Fragmented RNA was hybridized on a HumanHT-12 v4 Expression BeadChip (BD-103-0204, Illumina) by incubating at 58 C for 16 h. The hybridization mixture was detected with Cy3-Streptavidin (PA431001, GE Healthcare). The chips were washed, dried, and scanned using the Bead Array Reader (Illumina) and raw data were obtained using GenomeStudio soware V2011.1 (Illumina). All genes representing 2-fold dierence between the two groups were selected.

The experiments were performed according to the RayBio Human

Angiogenesis Antibody Array C series 1000 (AAH-ANG-1000, RayBiotech) guidelines. Briey, array A and B membranes were incubated with blocking buer for 30min. Aer removing the blocking buer, the membranes were incubated overnight with cell lysates in a humid chamber at 4C. The membranes were washed two times and incubated with the antibody cocktail for 2h. Aer the addition of HRP-conjugated streptavidin antibody for 2 h, the membranes were exposed to X-ray lm using the detection buers C and D. Intensity of each blot spot was quantied using Quantity One soware (Bio-Rad). The data were normalized using positive and negative controls.

About 20 nM FGF2 siRNA, FGF5 siRNA, and AccuTarget negative control siRNA (SN-1003, Bioneer) were transfected using lipofectamine 2000 transfection reagent (11668-019, Invitrogen). The transfected cells were incubated for 6 h in a transfection reagent mixture and maintained for 24 h in EGM-2 MV. Aer transfer to the 3D microuidic angiogenesis system, the cells were incubated for 48h and then harvested for the measurement of FGF2 and FGF5 mRNA. The FGF2 siRNA antisense sequence was as follows: 5-UAUACUGCCCAGUUCGUUUCAGUGC-3. FGF5 siRNA (M-011972-01-0005, Dharmacon) target sequences were as follows: 5-CAUAAGUUGUAUAGGCUAA-3, 5-CAACAAUAAGCCACGUCAA-3, 5-GCAAGUUCAGGGAGCGUUU-3, and 5-GUAUUGAAGUCACGUCAUU-3.

HUVECs and HAECs were washed twice with PBS, and lysis with 1mM phenylmethylsulfonyl uoride (P7626, Sigma) contained 1X cell lysis buer (9803, Cell signaling). Quantitative analysis of samples were determined with a Bradford assay dye reagent (500-0006, Bio-Rad). Fieen g of sample protein was mixed with 1X loading dye and boiled for 5min, and electrophoresis on 10% SDS-polyacrylamediume gel. Aer transferred to the polyvinylidene uoride membrane (ISEQ00010, Millipore), membranes were blocked in 5% skimmilk contained 1X TBST (WH400028806, 3M) for 1 hr. The membranes were incubated with anti-p-FGFR1 (1:1000, ab173305, abcam), anti-FGFR1 (1:1000, 9740, cell signaling) and anti-GAPDH (1:10000, G8795, Sigma) antibody at 4C for overnight. Next, the membranes were washed three times in TBST and incubated with a horseradish peroxidase-conjugated secondary anti-mouse HRP antibody (1:7000, SC-2005, Santa Cruz Biotechnology) and anti-rabbit HRP antibody (1:7000, SC-2030, Santa Cruz Biotechnology) in TBST at room temperature for 90 min. Chemiluminescence were visualized using Amersham ECL Prime Westerm Blotting Reagent (RPN2232SK, GE Healthcare Life Sciences) and exposed to X-ray lm. Quantication of blotting intensity were performed using Quantity One (Bio-Rad) program.

Data values are presented as the mean standard deviation (SD). Statistical signicance of the mean values was conrmed by Students t-test or ANOVA followed by Student-Newman-Keuls test. P < 0.05 indicates statistical signicance. All statistical analysis were performed using SigmaStat3.5 (SPSS, Chicago, IL).

SCIENTIFIC REPORTS

9

www.nature.com/scientificreports/

1. Morgan, C. & Nigam, Y. Naturally derived factors and their role in the promotion of angiogenesis for the healing of chronic wounds. Angiogenesis 16, 493502 (2013).

2. Rai, S. & Lyden, D. Therapeutic stem and progenitor cell transplantation for organ vascularization and regeneration. Nat. Med. 9, 702712, doi: 10.1038/nm0603-702 (2003).

3. Gridley, T. Notch signaling in vascular development and physiology. Development 134, 27092718 (2007).4. dela Paz, N. G., Walshe, T. E., Leach, L. L., Saint-Geniez, M. & DAmore, P. A. Role of shear-stress-induced VEGF expression in endothelial cell survival. J. Cell Sci. 125, 831843 (2012).

5. Neth, P., Nazari-Jahantigh, M., Schober, A. & Weber, C. MicroRNAs in ow-dependent vascular remodelling. Cardiovascular research 99, 294303 (2013).

6. Ingram, D. A. et al. Vessel wall-derived endothelial cells rapidly proliferate because they contain a complete hierarchy of endothelial progenitor cells. Blood 105, 27832786 (2005).

7. Coultas, L., Chawengsaksophak, K. & Rossant, J. Endothelial cells and VEGF in vascular development. Nature 438, 937945 (2005).8. Noseda, M. et al. Notch activation induces endothelial cell cycle arrest and participates in contact inhibition: role of p21Cip1 repression. Mol. Cell. Biol. 24, 88138822 (2004).

9. Gerhardt, H. VEGF and endothelial guidance in angiogenic sprouting. Organogenesis 4, 241246 (2008).10. Hellstrom, M. et al. Dll4 signalling through Notch1 regulates formation of tip cells during angiogenesis. Nature 445, 776780 (2007).11. Potente, M., Gerhardt, H. & Carmeliet, P. Basic and therapeutic aspects of angiogenesis. Cell 146, 873887 (2011).12. Barkefors, I. et al. Endothelial cell migration in stable gradients of vascular endothelial growth factor A and broblast growth factor

2: eects on chemotaxis and chemokinesis. J. Biol. Chem. 283, 1390513912 (2008).

13. Feldman, B., Poueymirou, W., Papaioannou, V. E., DeChiara, T. M. & Goldfarb, M. Requirement of FGF-4 for postimplantation mouse development. Science 267, 246249 (1995).

14. Floss, T., Arnold, H. H. & Braun, T. A role for FGF-6 in skeletal muscle regeneration. Genes Dev. 11, 20402051 (1997).15. Danilenko, D. M. Preclinical and early clinical development of keratinocyte growth factor, an epithelial-specic tissue growth factor. Toxicol. Pathol. 27, 6471 (1999).

16. Miller, D. L., Ortega, S., Bashayan, O., Basch, R. & Basilico, C. Compensation by broblast growth factor 1 (FGF1) does not account for the mild phenotypic defects observed in FGF2 null mice. Mol. Cell. Biol. 20, 22602268 (2000).

17. Izikki, M. et al. Endothelial-derived FGF2 contributes to the progression of pulmonary hypertension in humans and rodents. J. Clin. Invest. 119, 512523 (2009).

18. Przybylski, M. A review of the current research on the role of bFGF and VEGF in angiogenesis. Journal of wound care 18, 516519 (2009).

19. Lee, J. G. & Kay, E. P. FGF-2-induced wound healing in corneal endothelial cells requires Cdc42 activation and Rho inactivation through the phosphatidylinositol 3-kinase pathway. Investigative ophthalmology & visual science 47, 13761386 (2006).

20. Giordano, F. J. et al. Intracoronary gene transfer of broblast growth factor-5 increases blood ow and contractile function in an ischemic region of the heart. Nat. Med. 2, 534539 (1996).

21. Li, K., Stewart, D. J. & Ward, H. J. Technology evaluation: gene therapy (FGF-5), Vical. Current opinion in molecular therapeutics 1, 260265 (1999).

22. Hebert, J. M., Rosenquist, T., Gotz, J. & Martin, G. R. FGF5 as a regulator of the hair growth cycle: evidence from targeted and spontaneous mutations. Cell 78, 10171025 (1994).

23. Chung, S., Sudo, R., Zervantonakis, I. K., Rimchala, T. & Kamm, R. D. Surface-treatment-induced three-dimensional capillary morphogenesis in a microuidic platform. Advanced materials 21, 48634867 (2009).

24. Shin, Y. et al. Microuidic assay for simultaneous culture of multiple cell types on surfaces or within hydrogels. Nature protocols 7, 12471259 (2012).

25. Jeong, G. S. et al. Sprouting angiogenesis under a chemical gradient regulated by interactions with an endothelial monolayer in a microuidic platform. Analytical chemistry 83, 84548459 (2011).

26. Xiaozhen, D. et al. A novel in vitro angiogenesis model based on a microuidic device. Chin. Sci. Bull 56, 33013309 (2011).27. Hattori, K. et al. Microuidic perfusion culture chip providing dierent strengths of shear stress for analysis of vascular endothelial function. J. Biosci. Bioeng. 118, 327332 (2014).

28. Wong, K. H., Chan, J. M., Kamm, R. D. & Tien, J. Microuidic models of vascular functions. Annu. Rev. Biomed. Eng. 14, 205230 (2012).

29. Chung, S. et al. Cell migration into scaolds under co-culture conditions in a microuidic platform. Lab on a chip 9, 269275 (2009).30. Yamamura, N., Sudo, R., Ikeda, M. & Tanishita, K. Eects of the mechanical properties of collagen gel on the in vitro formation of microvessel networks by endothelial cells. Tissue Eng. 13, 14431453 (2007).

31. Bischel, L. L., Young, E. W., Mader, B. R. & Beebe, D. J. Tubeless microuidic angiogenesis assay with three-dimensional endothelial-lined microvessels. Biomaterials 34, 14711477 (2013).

32. Stroock, A. D. & Fischbach, C. Microuidic culture models of tumor angiogenesis. Tissue engineering. Part A 16, 21432146 (2010).33. Pek, Y. S., Wan, A. C. & Ying, J. Y. The eect of matrix stiness on mesenchymal stem cell dierentiation in a 3D thixotropic gel. Biomaterials 31, 385391 (2010).

34. Byeld, F. J. et al. Absence of lamin A prevents cells from responding to stiness gradients on gels coated with collagen but not bronectin. Biophys. J. 96, 50955102 (2009).

35. Kim, D. H., Provenzano, P. P., Smith, C. L. & Levchenko, A. Matrix nanotopography as a regulator of cell function. J. Cell Biol. 197, 351360 (2012).

36. Carmeliet, P. Mechanisms of angiogenesis and arteriogenesis. Nat. Med. 6, 389395 (2000).37. Risau, W. Mechanisms of angiogenesis. Nature 386, 671674 (1997).38. Azari, B. M. et al. Transcription and translation of human F11R gene are required for an initial step of atherogenesis induced by inammatory cytokines. Journal of translational medicine 9, 98 (2011).

39. Giannini, S. et al. Interaction with damaged vessel wall in vivo in humans induces platelets to express CD40L resulting in endothelial activation with no eect of aspirin intake. Am. J. Physiol. Heart Circ. Physiol. 300, H2072H2079 (2011).

40. Sung, H. J., Yee, A., Eskin, S. G. & McIntire, L. V. Cyclic strain and motion control produce opposite oxidative responses in two human endothelial cell types. Am. J. Physiol. Cell Physiol. 293, C87C94 (2007).

41. Aley, P. K. et al. Hypoxic modulation of ca(2+) signaling in human venous and arterial endothelial cells. J. Membr. Biol. 227, 151158 (2009).

42. Byrd, N. A. & Meyers, E. N. Loss of Gbx2 results in neural crest cell patterning and pharyngeal arch artery defects in the mouse embryo. Dev. Biol. 284, 233245 (2005).

43. Yu, Y. et al. Common variants near FRK/COL10A1 and VEGFA are associated with advanced age-related macular degeneration. Hum. Mol. Genet. 20, 36993709 (2011).

44. Maisonpierre, P. C. et al. Angiopoietin-2, a natural antagonist for Tie2 that disrupts in vivo angiogenesis. Science 277, 5560 (1997).45. Powers, C. J., McLeskey, S. W. & Wellstein, A. Fibroblast growth factors, their receptors and signaling. Endocrine-related cancer 7, 165197 (2000).

46. Zhang, X. et al. Receptor specicity of the broblast growth factor family. The complete mammalian FGF family. J. Biol. Chem. 281, 1569415700 (2006).

SCIENTIFIC REPORTS

10

www.nature.com/scientificreports/

47. Carmeliet, P. & Jain, R. K. Molecular mechanisms and clinical applications of angiogenesis. Nature 473, 298307 (2011).48. Moura, R. S., Coutinho-Borges, J. P., Pacheco, A. P., Damota, P. O. & Correia-Pinto, J. FGF signaling pathway in the developing chick lung: expression and inhibition studies. PloS one 6, e17660 (2011).

49. Murakami, M. et al. The FGF system has a key role in regulating vascular integrity. J. Clin. Invest. 118, 33553366 (2008).50. Schneider, C. A., Rasband, W. S. & Eliceiri, K. W. NIH Image to ImageJ: 25 years of image analysis. Nat. Methods 9, 671675 (2012).

This research was supported by Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Education (NRF-2015R1D1A1A01060397). Also, this research was supported by National Research Foundation of Korea (2014M3A7B4052193) funded by the Ministry of Science, ICT & Future Planning and the Human Resources Development of the Korea Institute of Energy Technology Evaluation and Planning (KETEP) grant funded by the Korea government Ministry of Trade, Industry & Energy (No. 20144010200770).

H.-R.S., H.E.J., H.J.J., S.-C.C., S.J.H., S.C. and D.-S.L. obtained the research funds and developed experimental protocol; H.-R.S., H.E.J., J.-H.K., C.-Y.P., J.-H.C. and L.-H.C. contributed to the acquisition and analysis of experimental data; H.-R.S. and H.E.J. draed the manuscript. All authors approved the manuscript.

Supplementary information accompanies this paper at http://www.nature.com/srep

Competing nancial interests: The authors declare no competing nancial interests.

How to cite this article: Seo, H.-R. et al. Intrinsic FGF2 and FGF5 promotes angiogenesis of human aortic endothelial cells in 3D microuidic angiogenesis system. Sci. Rep. 6, 28832; doi: 10.1038/srep28832 (2016).

This work is licensed under a Creative Commons Attribution 4.0 International License. The images or other third party material in this article are included in the articles Creative Commons license, unless indicated otherwise in the credit line; if the material is not included under the Creative Commons license, users will need to obtain permission from the license holder to reproduce the material. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/

SCIENTIFIC REPORTS

11